Transcript Evolution v. Revolution

Network Architecture (R02) #2
Evolution v. Revolution
Jon Crowcroft,
http://www.cl.cam.ac.uk/~jac22
http://www.cl.cam.ac.uk/teaching/1011/R02
Pros and Cons of IPv6 as it is now
Ipng requirements capture was
comprehensive
But decision was marred


1.
2.
3.
Famous false start:ISOCLNP==DECNET
S(simple|steve)IP == XNS
P(paul)IP == MPLS + Nimrod
Committee Compromise



Does mobile
Doesn’t do multihoming
Some other IPv6 plus points

Large addr space 




Internet of things (foreseen)
can hash into with low collision probability
Provider, geographic, topological
EID/Routing Hint separation (8+8)
Transport changes….


TCP socket = 5 tuple
Other transport use flow label?
Other IPv6 tricks


Could do zeroconf
Can use Host Id Protocol


or other crypto-assigned address mech,
Can in principle use flow label


Can use hop&end options without so much
impact on router fast path v. slow path for
IPv4 Options
including overloaded Router alert
Evolution v. Revolution

http://www.cl.cam.ac.uk/teaching/1011/R02/papers/p36-rexford.pdf



Is a local cached copy of counterpoint
between J Rexford (Princeton) and C
Dovrolis (Gatech) on evolution v. revolution
Prompted by discussion in GENI and other
programms
Clean Slate v. Incremental
Clean Slate

Throw out preconceptions/assumptions





Design new net
Then figure out how to deploy
Try and repeat 1973-1988 somehow
Useful at least in some Zen-like sense
What would you build for future with
nanotechnology, 3D printers, quantum
information processing etc?
Incremental




Sometimes term “incremental” used
“negatively”
Reality is its useful
Eg. Resource Pooling and Multipath IP
and Multipath TCP
See http://nrg.cs.ucl.ac.uk/mptcp/
Overlays and Underlays

IP was once an overlay




On x.25 and on PSTN
Then moved down to bare metal
ip on photons too
Overlays are a deployment/evolutionary
route

see national academy of science on “looking
over the fence at network research)
Other overlay schemes

Mbone, 6bone, abone





Typically tunnel…but


RON
Application Layer Multicast
IP on SMS and on Fb!
IP on DNS
IP in IP overhead
IP on VC - complexity



Circuit setup/teardown
#circuits, qos
routing
Underlays


more revolutionary than overlays
Not many examples 
MPLS, gMPLS


3gpp/cellular


Finesse management/routing problem
Finesse mobility&billing problem
Other?
Deployment of new transport

How can it be hard?




Some benefits of Multipath TCP




Because of firewalls & IDS and DPI
And packet normalisers
Default is “unknown” == “bad”
Resource pooling for multihomed server
Resilience to handover for multihomed mobile
Data center - removes hotspots in interconnect
Downside - general case needs multipath IP

Path selection from non multihomed end point?
Next talk for 19/10/10


What was really wrong with original
(Any Source) Multicast
How has one-to-many other ways - CDN
- been so succesful even though it
makes no efficiency sense?
First Essay (due Oct 29)




Take one of
LIRA/NIRA, I3, IPNL
Compare with baseline (IPv4/v6)
Evolution v. Revolution (see paper from CACM
on course web page)





What do they add to the mix
E.g. scale, multihome, multicast, mobile
E.g. security (what threats), performance
Where do they fall down
2-4 pages of notes

Remember to cite any sources you use!